For this module, I asked Claude AI a couple of silly questions about my astrological sign’s supposed love life. I plugged in my birthday information and kept adding more detail for a made-up, over-the-top dramatic romantic scenario to see how Claude would respond. I figured this would be funny to witness, but things quickly got creepy. Even as the questions I asked eventually turned away from astrology, Claude was insistent on providing me with astrologically inspired answers. As I kept feeding the bot more data, its language and tone became more “personable” and expressive, mimicking Gen Z phrases and overall demeaner. Using emojis and breaking grammar rules, it felt like I was having a text message conversation with another 22-year-old woman. While it didn’t surprise me that Claude just kept re-scrabbling the astrology information, what really scared me was how Claude quickly generated data that it thought would be most relatable to a 22-year-old female member of Gen Z. Once I gave it those basic facts it ran with them and crafted a “personality” that “catered” to the bot’s assumptions about me. This incident makes much more sense after reading, “The personhood trap: How AI fakes human personality” as AI uses pattern recognition within datasets to decode emotions and relationships, something in my opinion that are inherently human and therefore unmeasurable. Emotions are subjective and varied, therefore how could AI truly be able to decode and understand this quantity of variety?! (https://arstechnica.com/information-technology/2025/08/the-personhood-trap-how-ai-fakes-human-personality/). Even after reading all these articles and learning more about artificial intelligence, this technology still does not sit right with me. AI is treated as this all-purpose band-aid that will rapidly fix any problem just because it has amassed all this data across the internet. AI is just a collection of technological educated guesses based on mass pools of data on the internet, as discussed in Lee and Trott’s “jargon-free explanation” of AI. AI was making assumptions about me because of its predictions based on pooled data sources. (https://arstechnica.com/science/2023/07/a-jargon-free-explanation-of-how-ai-large-language-models-work/).
Knowing that AI is just some glorified magic 8-ball, I’m not surprised to learn about the harm AI’s “training process” causes the environment and even copyright. Ironically, AI bots are “trained” by humans as discussed in “AI Is a Lot of Work” (https://archive.is/20250614173310/https://www.theverge.com/features/23764584/ai-artificial-intelligence-data-notation-labor-scale-surge-remotasks-openai-chatbots). Reading about the amount of energy, electricity, and water AI requires is pure insanity. The infrastructure supporting AI is a continuous, harmful, and resource-draining system. AI’s extreme electricity needs creates additional carbon dioxide emissions and strains electrical grides, while the amount of water required to cool down the mechanical parts behind the scenes harms the environment. Additionally, the data centers where AI is managed use up considerable electricity. (https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117). The training requirements for AI really gets it into trouble as seen with its impact on the environment as well as its legal ramifications. The AI company Anthropic, responsible for Claude, downloaded pirated books off the internet to “train” the AI bots and were taken to court. After losing the case, Anthropic paid a large settlement amount to the authors and publishing companies that had had their copyrighted material used without permission. While this case helped define what AI shouldn’t do, it also defined what it can do. AI companies have followed fair use laws more closely now as these organizations have started to buy licensing rights to copyrighted content which is then used to train AI (https://www.removepaywall.com/search?url=https://www.nytimes.com/2025/09/05/technology/anthropic-settlement-copyright-ai.html).
Advocating for the idea that libraries can act as the trainers of AI, libraries and librarians are considering how they can counteract the buying up of creative content by AI companies. Arguing that libraries and librarians already have a variety of literary expertise, experience in copyright law, and specialty in creatively informed research, libraries do seem like more ethical AI trainers in comparison to tech companies like Meta, Amazon, and Google who are more concerned with consolidating a wealth of data (https://newsletter.dancohen.org/archive/humane-ingenuity-53-books-are-big-ais-achilles-heel/).
The more I write this blog post, the more I dislike AI and I’m sorry to say that I have very few positive things to say about it. One positive attribute of AI that did stand out to me was discussed in Sonja Anderson’s “See Newly Discovered Nazca Drawings That Depict Llamas, Human Sacrifices and More.” Archaeologists trained AI to help identity vague geoglyphs from the ancient Nazca culture. While archaeologists informed and checked AI’s work, the AI bot was extremely helpful and successful in identifying new geoglyphs, so much so that the archaeologists hope to continue to use it within their future work. A key point here is that the human archaeologists checked the AI’s input and output. While every single article cited for this module discusses the need for AI and humans to work together, Anderson’s article finally helped me see that there can be a “checks and balances” system between AI and human thought (even though personally speaking I’m not a fan of this idea). (https://www.smithsonianmag.com/smart-news/see-newly-discovered-nazca-drawings-depict-llamas-human-sacrifices-more-180985133/; https://dhlab.hypotheses.org/6060).
I really liked Piers Gelly’s article that discusses the role of AI in his college level English classroom. Running with his use of the metaphor of considering AI like the calculator of English, this made me think about my own usage of a calculator. I’ll be the first to admit that I can’t do any advanced mental math in my head anymore because I have become so reliant on the conveyance of my iPhone calculator. We must learn how to do math without the calculator, and we must learn to write without the help of AI as ultimately, we can become too reliant on these tools. AI, like a calculator, can be helpful for basic writing tasks like editing but it cannot write the essay or solve the entire math problem for you as Gelly conveys throughout his article (https://lithub.com/what-happened-when-i-tried-to-replace-myself-with-chatgpt-in-my-english-classroom/).
Hey Ally, great post. Like you, I am also concerned with how “lifelike” AI writing is getting. By making the algorithm produce language similar to the audience, it is almost making it more “likable” for general audiences. The more AI is used, the more damage it can cause in the long run, such as fake sources and images. I don’t think I will use AI in the future due to the many issues inherent to it.